Specto: A Three Day Challenge By Swift UI Jam
Specto is an audio visualizer application that also transcribes user input using Apple speech recognition API. Each recording session is tagged with highlighted keywords by the TextRank algorithm. Designed and developed by Taha Mousavi, Abbas Mousavi and Moeen Zamani for SwiftUI Jam, Feb 2021.
In 3 days, I worked on Finding creative ways to visualize recorded voice and naming the recorded voice files.
Visualization
For visualization, I suggested using the FFT method for detecting frequencies and using that for creating a pattern on a vinyl disc. After several tests and improvements, Abbas and Moeen found an optimum and beautiful method to implement it.
We uses angular gradients to project the FFT to concentric circle paths. The whole image shows 50 FFTs calculated five times per second each with 33 components which makes 1650 visual components in total. We started with multiple curves in different colors but that implementation was slow, so we switched to angular gradients producing smoother graphics with much better performance.
Here are the final results for different voice recorded.
Prototypes
Interactive prototypes by Principle.
Naming of files
For naming the recorded file, We use highlighted keywords by the TextRank algorithm. By using different fonts and colors, Specto assigns three words to every file and creates an album cover for them.
App Icon
App icon inspired by 45 rpm adapters.
